AI August 2017

ai@lists.wikimedia.org

5 participants
5 discussions

More/better model information and "threshold optimizations
by zppix e 31 Aug '17

31 Aug '17

We'd like to let you know about an upcoming change to ORES that might break your code. We've made some improvements to the way that "model_info" is reported from ORES. Rather than us needing to specify thresholds when we train a model (e.g. "filter_rate_at_recall(min_recall=0.9)"), the system now allows arbitrary queries (e.g. "filter_rate @ recall >= 0.9"). This if more flexible but it may cause problems for anyone who was using the machine-readable thresholds to support a UI in the past. Scores themselves will remain unchanged. The model info and threshold systems will stay in the current state until we are ready to deploy the newer version to production (probably around Sept. 20th). If you have any questions, please feel free to contact the Scoring Platform Team, via email (ai(a)lists.wikimedia.net) or IRC (chat.freenode.net:6668 #wikimedia-ai) -- Zppix Volunteer developer for Wikimedia Sent from my iPhone

2 2

Wikilabels incident: Reversed diffs
by Amir Tafreshi 31 Aug '17

31 Aug '17

Hello, Today, we discovered a major regression in Wikilabels. We've patched the issue and made an emergency deployment. We also deleted some labels that were saved while the system was compromised. Here is the details of of what happened: https://phabricator.wikimedia.org/phame/post/view/69/wikilabels_incident_re… We're very sorry for this inconvenience, Yours truly -- Amir Sarabadani Tafreshi Software Engineer (contractor) ------------------------------------- Wikimedia Deutschland e.V. | Tempelhofer Ufer 23-24 | 10963 Berlin http://wikimedia.de Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

1 0

[Deadline Extended] Call for paper of the 22nd Conference on Technologies and Applications of Artificial Intelligence (TAAI 2017)
by vincenttsai＠iis.sinica.edu.tw 30 Aug '17

30 Aug '17

Dear all, The submission deadline of TAAI 2017 is extended to Sep. 11, 2017. The 22nd Conference on Technologies and Applications of Artificial Intelligence (TAAI 2017) will be held in National Taiwan University, Taipei, Taiwan on December 1-3 2017. *International Call For Papers* The 2017 Conference on Technologies and Applications of Artificial Intelligence (TAAI 2017) is one of the most important annual academic meetings on Artificial Intelligence in Taiwan. The conference in 2017 will focus on the following theme: Achieving excellence through Artificial Intelligence Within the scope of this theme we are seeking paper submissions to address the following topics of particular interest include, but are not limited to: - Agents - AI Applications - AI Architectures - Automated Reasoning - Cloud Computing in AI - Cognitive Modeling - Computer Games - Intelligent Environments - Intelligent Human-Computer Interaction - Knowledge-Based Systems - Knowledge Representation - Logics in AI - Machine Learning - Mobile Intelligence - Computer Supported Collaborative Learning & Personalized Learning - Natural Language Processing - Pattern Recognition - Data Science - Planning - Computer Vision - Data Mining - Distributed AI - Evolutionary Computation & Genetic Algorithms - Probabilistic and Uncertain Reasoning - Problem Solving and Search - Robotics - Fuzzy Systems - Hybrid Systems - Information Retrieval - Semantic Web - Social Computing - Speech Recognition and Synthesis - Intelligent E-Learning & Tutoring - Ubiquitous Intelligent Applications - Web Intelligence - Big Data Analysis *IMPORTANT DATES* Submission deadline Sep. 11, 2017 (Extended) Acceptance notification Oct. 07, 2017 Camera ready Oct. 21, 2017 Conference Date December 1-3, 2017 *Submission Information & Guidelines* The TAAI 2017 paper template shall be used to prepare your full papers in 6 pages and short papers in 4 pages(including graphics and references). The paper format should follow IEEE Manuscript Templates for Conference Proceedings(http://www.ieee.org/conferences_events/conferences/publishing/t… to prepare manuscripts. The papers shall be submitted in English (CMT: https://cmt3.research.microsoft.com/User/Login?ReturnUrl=%2FTAAI2017) and shall describe original work that has not previously been presented at TAAI. For more detailed information, please refer: https://www.csie.ntu.edu.tw/~taai2017/cfp.html -- 蔡睿翊 Vincent Jui-Yi, Tsai 中央研究院資訊科學研究所 Institute of Information Science (IIS), Academia Sinica Tel: (02)27883799 #1371 Email: vincenttsai(a)iis.sinica.edu.tw ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.

1 0

ML in the browser
by Toby Negrin 16 Aug '17

16 Aug '17

H/T to Magnus https://research.googleblog.com/2017/08/harness-power-of-machine-learning-i… -Toby -- -Toby

1 0

[RfC] [ORES] Do you use reverted model for wikis that have damaging ones?
by Amir Tafreshi 03 Aug '17

03 Aug '17

Hey, For patrolling work, ORES usually has two levels of support: - For basic support we usually provide a model that is called 'reverted' and has less accuracy. It also risks perpetuating editor biases due to the lack of differentiation between reasons that a change may have been reverted. - For advanced support, we require manual labeling of a large sample of edits, but then we can provide two more models: 'damaging' and 'goodfaith'. ORES review tool can only be enabled on wikis where advanced support is available and most other tools prefer the 'damaging' over the basic 'reverted' model as well. So, for performance and capacity reasons we think it makes sense to remove 'reverted' models from the ORES service when 'damaging' model is made available. However, we also want to be careful about making sure this change doesn't disrupt the work of tool developers that make use of the ORES service. If you do, please voice your concerns now. If there is no objection within the next two weeks, we'll begin the process of removing the 'reverted' model for wikis that have the 'damaging' model available. Related phabricator card: https://phabricator.wikimedia.org/T171059 Best -- Amir Sarabadani Tafreshi Software Engineer (contractor) ------------------------------------- Wikimedia Deutschland e.V. | Tempelhofer Ufer 23-24 | 10963 Berlin http://wikimedia.de Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

1 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

AI August 2017